fork of https://github.com/sourcegraph/zoekt
indexserver: truncate failure messages in status updates (#1070)
Truncate failure messages to 12 KiB before sending index status updates to Sourcegraph. This limit is modeled after Kubernetes, which similarly truncates termination messages on the assumption that complete failure details are available in container logs.
author
Stefan Hengl
committer
GitHub
date
(Jun 5, 2026, 2:22 PM +0200)
commit
893a5238
893a523804f1bbb7132a65b2cc5de348cf4cd2c2
parent
244807b6
244807b64cb5844c48cc6a93fe197519e3ec6e52
+74
-1
+52
-1
cmd/zoekt-sourcegraph-indexserver/main.go
+52
-1
cmd/zoekt-sourcegraph-indexserver/main.go
···
44
44
"sync"
45
45
"text/tabwriter"
46
46
"time"
47
+
"unicode/utf8"
47
48
48
49
grpcprom "github.com/grpc-ecosystem/go-grpc-middleware/providers/prometheus"
49
50
"github.com/grpc-ecosystem/go-grpc-middleware/v2/interceptors/retry"
···
254
255
// time." famous last words. A client was indexing a monorepo with 42
255
256
// cores... 5m was not enough.
256
257
const noOutputTimeout = 30 * time.Minute
258
+
259
+
const (
260
+
maxFailureMessageBytes = 12 * 1024
261
+
failureMessageTruncationMarker = "\n\n[Error message truncated]\n\n"
262
+
)
257
263
258
264
func (s *Server) loggedRun(tr trace.Trace, cmd *exec.Cmd) (err error) {
259
265
out := &synchronizedBuffer{}
···
707
713
var indexTimeUnix int64
708
714
if indexErr != nil {
709
715
state = configv1.UpdateIndexStatusRequest_Repository_STATE_FAILURE
710
-
failureMessage = indexErr.Error()
716
+
failureMessage = truncateFailureMessageForSourcegraph(indexErr.Error())
711
717
712
718
// On failure, metadata may not exist yet. Include index time if we can,
713
719
// but do not block reporting the failure status.
···
739
745
}
740
746
741
747
return nil
748
+
}
749
+
750
+
func truncateFailureMessageForSourcegraph(s string) string {
751
+
if len(s) <= maxFailureMessageBytes {
752
+
return s
753
+
}
754
+
755
+
budget := maxFailureMessageBytes - len(failureMessageTruncationMarker)
756
+
headBytes := budget / 2
757
+
tailBytes := budget - headBytes
758
+
759
+
return utf8PrefixBytes(s, headBytes) + failureMessageTruncationMarker + utf8SuffixBytes(s, tailBytes)
760
+
}
761
+
762
+
func utf8PrefixBytes(s string, maxBytes int) string {
763
+
if len(s) <= maxBytes {
764
+
return s
765
+
}
766
+
767
+
if maxBytes <= 0 {
768
+
return ""
769
+
}
770
+
771
+
for maxBytes > 0 && !utf8.RuneStart(s[maxBytes]) {
772
+
maxBytes--
773
+
}
774
+
775
+
return s[:maxBytes]
776
+
}
777
+
778
+
func utf8SuffixBytes(s string, maxBytes int) string {
779
+
if len(s) <= maxBytes {
780
+
return s
781
+
}
782
+
783
+
if maxBytes <= 0 {
784
+
return ""
785
+
}
786
+
787
+
start := len(s) - maxBytes
788
+
for start < len(s) && !utf8.RuneStart(s[start]) {
789
+
start++
790
+
}
791
+
792
+
return s[start:]
742
793
}
743
794
744
795
func sglogBranches(key string, branches []zoekt.RepositoryBranch) sglog.Field {
+22
cmd/zoekt-sourcegraph-indexserver/main_test.go
+22
cmd/zoekt-sourcegraph-indexserver/main_test.go
···
12
12
"slices"
13
13
"strings"
14
14
"testing"
15
+
"unicode/utf8"
15
16
16
17
"github.com/google/go-cmp/cmp"
17
18
sglog "github.com/sourcegraph/log"
···
55
56
s := &Server{}
56
57
_, err := s.index(context.Background(), &indexArgs{})
57
58
require.ErrorIs(t, err, tenant.ErrMissingTenant)
59
+
}
60
+
61
+
func TestTruncateFailureMessageForSourcegraph(t *testing.T) {
62
+
t.Run("preserves short message", func(t *testing.T) {
63
+
require.Equal(t, "boom", truncateFailureMessageForSourcegraph("boom"))
64
+
})
65
+
66
+
t.Run("truncates oversized utf8 message", func(t *testing.T) {
67
+
input := strings.Repeat("é", maxFailureMessageBytes) + "tail"
68
+
69
+
got := truncateFailureMessageForSourcegraph(input)
70
+
parts := strings.Split(got, failureMessageTruncationMarker)
71
+
72
+
require.Len(t, parts, 2)
73
+
require.LessOrEqual(t, len(got), maxFailureMessageBytes)
74
+
require.True(t, utf8.ValidString(got))
75
+
require.NotEmpty(t, parts[0])
76
+
require.True(t, strings.HasPrefix(input, parts[0]))
77
+
require.Contains(t, parts[1], "tail")
78
+
require.Less(t, len(parts[0])+len(parts[1]), len(input))
79
+
})
58
80
}
59
81
60
82
func TestServer_parallelism(t *testing.T) {